Benchmarking of computational error-correction methods for next-generation sequencing data

This project contains the links to the datasets and the code that was used for our study : "Benchmarking of computational error-correction methods for next-generation sequencing data"

Table of contents

• How to cite this study
• Reproducinf results
  • Tools
  • Data
  • Notebooks and Figures
• License
• Contact

How to cite this study

Mitchell, Keith, et al. "Benchmarking of computational error-correction methods for next-generation sequencing data" bioRxiv, doi: https://doi.org/10.1101/642843

Reproducing results

Tools

We have evaluated 10 error correction tools: BFC, Bless, Coral, Fiona, Lighter, Musket, Pollux, Reckoner, Racer and SGA. Details about the tools and instructions for running can be found in our "paper".

We have prepared "wrappers" in order to run each of the respective tools as well as create standardized log files.

We have also prepared "scripts" to perform the evaluation of the error correction methods.

Data

We have evaluated 5 datasets composed of raw reads and their respective true reads (gold standard).

• D1 dataset: D1 was produced by computational simulations using a customized version of the tool WgSim. We generated simulated data mimicking the WGS human data using a customized version of the tool WgSim. Read coverage varied between 1 and 32. The WgSim fork is available at https://github.com/mandricigor/wgsim.

• D2 dataset: Raw reads corresponding to 8 samples (SRR1543964, SRR1543965, SRR1543966, SRR1543967, SRR1543968, SRR1543969, SRR1543970, and SRR1543971) were downloaded from https://www.ncbi.nlm.nih.gov/. The error-free (true) reads for the D2 dataset were generated using a UMI-based high-fidelity sequencing protocol, also known as safe-SeqS.

• D3 dataset: We generated simulated data mimicking the TCR-Seq data using the T cell receptor alpha chain (TCRA). Samples have read lengths of 100bp and read coverage varied between 1 and 32.

• D4 dataset: D4 corresponds to HIV population sequencing of an infected patient. The error-free (true) reads for the D4 dataset were generated using a UMI-based high-fidelity sequencing protocol.

• D5 dataset: We prepared the viral dataset D5 using real sequencing data from NCBI with the accession number SRR961514. Each read was assigned to the reference with which it has a minimum number of mismatches. The original error rate in the dataset was 1.44%. We modified these reads as follows: first, we corrected the corresponding portion of errors with a corresponding reference nucleotides to obtain different levels of errors in the datasets (1.44%, 0.33%, 0.1%, 0.033%, 0.01% , 0.0033%, 0.001%, 0.00033%, 0.0001%); We also created datasets with mixtures of two haplotypes with the original 1.44% error rate but with different levels of diversity between haplotypes (Hamming distance=5.94%, 0.29%, 0.02%). We applied a haplotype-based error correction protocol to eliminate sequencing errors from the D5 dataset.

Notebooks and Figures

We have prepared Jupyter Notebooks that utilize the raw data described above to reproduce the results and figures presented in our manuscript.

• D1 dataset (E.coli) Jupyter Notebook
• D1 dataset (Human) Jupyter Notebook
• D2 dataset Jupyter Notebook
• D3 dataset Jupyter Notebook
• D4 dataset Jupyter Notebook
• D5 dataset Jupyter Notebook

License

This repository is under MIT license. For more information, please read our LICENSE.md file.

Contact

Please do not hesitate to contact us (mangul@usc.edu) if you have any comments, suggestions, or clarification requests regarding the study or if you would like to contribute to this resource.